Yasuhiro Fujita

Use uppercase letters in hostnames

5

When I put an entry with a hostname that contains uppercase letters to `.ssh/assh.yml`, ``` hosts: SomeHost: User: someuser ``` it always gets lowered in `.ssh/config`. ``` Host somehost User...

Some evaluation results are missing

2

In `scores.txt` of the current uploaded trained model, evaluation results at `55000000` and `56000000` are missing. https://github.com/muupan/async-rl/blob/0ec501c36b6cdbe3888d3a6fc0043e4bc6c2cba3/trained_model/breakout/scores.txt#L55 I don't know why and whether it can affect performance. I need to...

Even after saving the final model `80000000_finish.h5`, some processes continue running. This issue is maybe related to #5 .

Not sample efficient enough

4

From Figure 6 in the paper, their A3C only needs 20 epochs (20 million steps) to achieve average scores of around 400 at Breakout. My current implementation needs more.

Too many deep RL papers to list up!

2

Now I find it hard to naively classify all the papers and manually list them up. I need a better way to maintain this list.

Make sure MJCFBaseBulletEnv.render returns a uint8 array

1

The issue is `MJCFBaseBulletEnv.render` returns an `int64` array. Here is how to reproduce the issue. ```python import gym import pybullet_envs env = gym.make("AntBulletEnv-v0") env.reset() array = env.render(mode="rgb_array") print(array.dtype) ``` What...

Reproducing the scores reported by the IQN paper

29

Thank you for opensourcing such a great code! I have questions about your IQN implementation, especially on how it can reproduce the scores reported by th paper. First, your config...

Specify successful configurations for examples

1

Current examples don't specify in what configuration they work well, except newer ones (train_pcl_gym.py and train_reinforce_gym.py). Such instructions are important because users can easily confirm that the implementations actually work....

~Merge #436 first.~